Journal of Chromatography B, 815 (2005) 39–50

Review

Strategies for revealing lower abundance proteins in two-dimensional

protein maps
Nuzhat Ahmed∗ , Gregory E. Rice
Gynaecological Cancer Research Centre, Royal Women’s Hospital, Department of Obstetrics and Gynaecology, The University of Melbourne,
132 Grattan Street, Carlton, Vic. 3053, Translational Proteomics Division, The Baker Medical Research Institute,
75 Commercial Road, Melbourne Australia

Received 7 June 2004; accepted 22 October 2004

Available online 28 November 2004

Abstract

One of the most challenging contemporary research endeavors is the mapping of proteins and establishing their linkages to normal and
pathological conditions. The availability of current proteomics technologies has greatly facilitated the separation and identification of proteins
in a complex protein mixture by standard two-dimensional gel electrophoresis and subsequent MALDI-TOF mass spectrometry. Due to the
huge differences in the distribution of proteins in complex proteomes of humans, the detection and identification of proteins expressed in
low copy number is a major challenge. The low abundance of important physiologically relevant proteins has rendered their analyses almost
impossible without some means of prior purification and enrichment from tissue lysates or biological fluids. It is the current limits of detection
of the methods that are used that prevents the detection of these proteins not the proteins themselves. More importantly, considering the
frequency at which post-translational modifications of proteins occur, the separation of protein isoforms is essential to understand biological
changes, and two-dimensional gel electrophoresis remains the only technique that can offer sufficient resolution to address this issue at a
functional level. Cellular fractionation techniques followed by specific affinity probes for tracking target proteins have been developed to
deplete the proteome of high abundance proteins in order to increase the sample loading for achieving greater sensitivity for proteins present
in low abundance. Those applications can entail the removal of one protein or a class of proteins that interferes with the resolution of proteins
in a 2-DE map. Moreover, the use of better solubilizing detergents in combination with an overlapping narrow immobilized pH gradients,
results in higher resolution by stretching the protein pattern in the first dimension. In this review we will discuss strategies to remove high
abundance proteins that can result in the visualization and detection of low abundance proteins in biological samples. The potential use of
these strategies, as a means of developing diagnostic tools for early screening of diseases and identification of drug targets for therapeutic
intervention, will also be discussed.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Low abundance protein; MALDI-TOF; Gel electrophoresis

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2. Strategies for the enrichment and identification of low abundance proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3. Depletion of high abundance protein in tissue homogenates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.1. Sample preparation and selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.2. Enrichment of low abundance proteins by affinity methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.3. Low abundance protein enrichment by laser micro-dissection and other mechanical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

∗ Corresponding author. Tel.: +61 3 9344 2616; fax: +61 3 9344 2619.
E-mail address: nuzhata@unimelb.edu.au (N. Ahmed).

1570-0232/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jchromb.2004.10.070
40 N. Ahmed, G.E. Rice / J. Chromatogr. B 815 (2005) 39–50

3.4. Enrichment of low abundance protein by sub-cellular fractionation method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

3.5. Affinity chromatography for the enrichment of low abundance protein . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.6. Tracking low abundance cell signaling proteins using proteomics approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.7. Enrichment of low abundance hydrophobic and basic proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4. Strategies to enrich low abundance protein in biological fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.1. Application of affinity chromatography in biological fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2. High abundant protein depletion strategies using multiple chicken IgY antibodies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.3. Low abundance protein enrichment in biological fluids by sample pre-fractionation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.4. Use of overlapping narrower pI range strips for enrichment of low abundance protein in biological fluids . . . . . . . . . . . . . . . . 46
4.5. Enrichment of low abundance protein by localized venous drainage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5. Advances and limitations of 2-DE based technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
6. Future perspectives and conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

1. Introduction sion of different proteins in a proteome, modifications and

interaction of proteins that occur due to changes in the pro-
The study of human proteomes, their diversity and their teome during the developmental process, disease-state or
relationship to genome has raised great interest. The Hu- exposure to an external stimuli. The core element of pro-
man Genome project provides evidence for approximately teomics analysis is to combine the separation of proteins
30,000–40,000 genes in the human proteome [1], a range in a two-dimensional map together with mass spectrome-
that is not much in excess of the gene numbers in the fruit try (MS) or tandem MS (MS–MS) protein identification. The
fly Drosophila melanogaster. The seemingly small number major advantage of this technique is that it enables the simul-
of genes in the human genome compared to other less com- taneous separation, visualization and identification of hun-
plex organisms has evoked scientific debate about the extent dreds of unknown proteins at different modification states.
of complexity and diversification of the human genome and No other method is able to achieve this at the present time.
its differences with other organisms such as bacteria, yeast, The technology has successfully been applied to gain under-
flies, other mammals, etc. It is apparent that the paradigm standing of the protein profile of simple organisms such as
of ‘one gene encoding a single protein’ is no longer appli- M. genitalium [5], E. coli [6], yeast [6], etc. Characterizing
cable because of differential RNA processing such as alter- the proteome of a complex organism such as humans how-
native RNA splicing, trans-slicing RNA events, overlapping ever, challenges the limitations of currently available tech-
transcription events, etc. [2]. The end result of these pro- nologies.
cesses includes post-translational protein modifications re- One of the challenges in identifying proteins in a com-
sulting in multiple protein products for a single encoded plex proteome such as human is the presence of proteins in
gene. Hence, for every eukaryotic genome studied to date, wide dynamic concentration ranges from very high levels
the actual number of distinct proteins formed in the differ- for albumin to low for a specific hormone or a regulatory
ing proteomes during the development far exceeds the num- protein. The differences in the concentration of such pro-
ber of transcription units that encoded them. Additionally, teins may be over thousands to a million times. Hence, a
protein degradation and turnover can significantly influence small sample volume (about 10–100 ␮l) usually applied for
the intracellular concentration of active protein molecules. It proteomics analysis, results in the acquisition of data domi-
has been estimated, that the average number of proteins per nated by high abundance proteins leaving a large percentage
gene is one or two in bacteria, three in yeast and three to of the expressed proteins with insufficient quantities unde-
more than six in human [3]. Hence, the extent of diversity tected. This large percentage of protein expressed in lower
and complexity resulting from post-translational modifica- concentration constitutes of low abundance protein and in-
tion and degradation is tremendous and can only be under- cludes proteins required for important regulatory processes
stood by qualitative and quantitative analysis of gene expres- or may represent proteins with potential biological marker
sion at the level of functional protein. Therefore, a direct possibility or may be a likely target for drug intervention.
measurement of protein expression in different proteomes To detect these low abundance proteins of greater biolog-
is essential to analyze and understand biological processes ical importance there is an urgent need to develop meth-
during development, in disease and in normal conditions. ods that will remove and deplete the relevant proteome of
The interface between protein biochemistry and molecular high abundance proteins, resulting in the analysis of 3–5-
biology for the global analysis of gene expression and sub- fold more proteins present in lower concentration. Strate-
sequent post-translational modification of product per se is gies to effectively remove high abundance proteins to iden-
termed ‘proteomics’ [4]. Proteomics is a technology-based tify proteins of low abundance therefore have gained much
science that studies in a high-throughput mode the expres- interest.
 N. Ahmed, G.E. Rice / J. Chromatogr. B 815 (2005) 39–50
2. Strategies for the enrichment and identification of
low abundance proteins
As discussed above, proteomics is an area of research that
seeks to define the function and relative expression profiles
of proteins encoded by a given genome at a given time in
a given cellular location. The technology separates, identi-
fies, and characterizes the proteins expressed, retained, se-
creted or released by a cell or tissue in order to establish their
function(s) and potential relationship during developmental
phase and or onset or progression of diseases, as well as re-
lapse and/or response to therapy. The first step for proteomic
analysis requires an adequate sample preparation that will
enable visualization and identification of relevant proteins of
low abundance. The process is equally important for biolog-
ical samples such as cell lysates as well as biological fluids
such as serum, plasma, saliva, cerebospinal fluids, uterine
washings, amniotic fluids, urine, etc. Several factors have an
impact on the detection of low abundance proteins in these
samples. The first is the dynamic range of protein concen-
tration in a sample. A distinction must be made between the
dynamic range of protein concentration in a cell and in bi-
ological fluids such as plasma. Corthals et al. [7] predicts
that the dynamic range of protein concentration in plasma
could be 12 orders of magnitude as the presence of albumin
represents over 50% of total plasma protein. In a cell, this
situation is less extreme and the most abundant proteins in
yeast represent 4% of the total protein [8]. Hence, strategies
for low abundance protein separation in cellular situation are
slightly different to that applied to biological fluids. As con-
tribution of high abundance proteins in biological fluids is
greater than that in tissues or cell preparations, visualization
and detection of biologically/clinically low abundance pro-
teins in biological fluids mandates the selective removal of
high abundance proteins. This review will focus separately on
methods applied for the removal of high abundance proteins
in cells/tissue homogenates as well as biological fluids.
3. Depletion of high abundance protein in tissue
homogenates
3.1. Sample preparation and selection
An optimized sample preparation procedure for the visu-
alization of low abundance protein in tissue samples includes
preparation and purification of the cell type of interest. This
will result in optimal cell lysis and solubilization of protein
to gain optimal yield of low abundance protein. Sample se-
lection is of paramount importance, for example use of fresh
samples for 2-DE analysis was reported to be better than
working with frozen samples [9]. Prolonged storage of sam-
ples may result in the modification of proteins resulting from
storage leading to misleading results [10]. Another challenge
to the analysis of tissue proteomes is tissue heterogeneity. The
sample to be analyzed should be pure and relevant. For exam-
41
ple, in case of cancer proteome analysis, the sample should
be free of stroma, blood, serum, etc and represent only tumor.
In that case, the use of cancer cell lines derived from tumors
is ideal but the representation of the cell line model to the in
vivo situation and the transfer of in vitro results to in vivo
conditions are still debatable.
3.2. Enrichment of low abundance proteins by affinity
methods
Different methods have been developed to deplete tissue
samples of non-relevant cell types in order to enrich the tissue
homogenate of relevant low abundant protein. The separation
of epithelial cells using Dynabeads from colon crypts iso-
lated by mechanical preparation has been shown [11]. Using
such techniques significant changes in the protein expression
profile between normal mucosa and colorectal cancer were
observed using 2-DE gels [11]. Others have used antibody-
based purification strategies to purify cells of interest in whole
tissue [12]. Other investigators have used tissue homogenates
prepared from pieces of tumor without purification of tumor
cells for 2-DE analysis [13]. As stromal components are part
of a tumor and many genes are differentially expressed in
stroma related to the tumor, the usage of whole tumor tis-
sues without purification is justified. In this context, one can
consider the example of metalloproteinases that are often ex-
pressed by tumor surrounding stroma but not by tumor cells
[14]. Solublization of protein in such mixed tissue lysates,
however, can be cumbersome and can result in complex pat-
terns due to the presence of several cell types making the
interpretation of proteomics study difficult.
Another method of determining low abundance protein
in biological samples is radioisotope labeling of cells be-
fore 2-DE analysis. Such labeling technique can be per-
formed by 35 S-methionine or 32 P-phosphorus [15] labeling
of cells in culture but it may not always be convenient to
radioactively label proteins in tissues. Radioisotopic label-
ing techniques provide a quantitative measure of a protein
spot in response to external stimuli or different physiological
conditions.
3.3. Low abundance protein enrichment by laser
micro-dissection and other mechanical methods
Different methods have been developed to deplete tumor
cells of the surrounding medium. Recently, the laser capture
micro-dissection (LCM) technology has gained prominence
in separating defined populations of cells in a malignant tis-
sue sample [16]. This method of sample preparation has been
successful in prostate cancer and has resulted in preparation
of samples with high tissue heterogeneity [17]. A compar-
ative analysis of 2-DE profile obtained from whole tissues
and LCM has recently been shown [18]. A high degree of
similarity between the different samples but with more en-
richment of some proteins in LCM derived samples was ob-
served. Other groups with different malignancies reported
42 N. Ahmed, G.E. Rice / J. Chromatogr. B 815 (2005) 39–50
similar result [19–21]. Even though the LCM technique of-
fers benefits for protein profiling of a homogenous population
of cells the technique has several drawbacks. The technique
is low throughput as only small number of samples can be
processed at a time. Moreover, sample misappropriation due
to misleading sample staining can be a problem. Hence, a
high quality of dissected material can only be guaranteed if
the LCM is performed by or under the supervision of a trained
pathologist.
Another method of collecting tumor cells of the same en-
tity is to perform mechanical cell extraction either by fine
needle aspiration (thyroid and breast tumors) [22] or by sur-
face scraping of tumor tissues (prostate and colon tumors)
[23]. As the tumor cells are loosely attached to the connec-
tive tissue and will be preferentially released by mechanical
force the method generally results in pure tumor cell popu-
lation.
3.4. Enrichment of low abundance protein by
sub-cellular fractionation method
For the enrichment of low abundance proteins in biolog-
ical sample biochemical protein-enriching approaches are
used. The original protein mixture is separated into fractions
of different cellular organelles by density-dependent ultra-
centrifugation procedures resulting in fractions enriched in
organelle-specific low abundance protein. This approach im-
proves resolution and enables one to increase the number
of protein spots that can be resolved and identified with the
probability of identifying low abundant proteins associated
with a particular organelle. Such techniques have been suc-
cessfully applied to identify low abundant membrane protein
[24]. However, such fractionation procedures require larger
quantities of biological samples, which in all cases is not
always possible.
Further enrichment of low abundant proteins separated
into organelle fractions can be achieved by selective frac-
tionation, chromatography or electrophoretic procedure [25].
Preparative electrophoresis is a general method of protein
purification. The electrophoretic method may comprise sep-
arating protein mixtures by preparative electrophoresis on
the basis of protein size, usually in the presence of ionic
detergents (Prep Cell, Bio-Rad Laboratories) [26], or by iso-
electrofocusing on the basis of protein charge either in the
presence of ampholines (Rotofor system, Bio-Rad Labora-
tories) or with the use of multi-compartment electrolyzers
with isoelectric immobilized pH gradient (IPG) membranes
[27].
3.5. Affinity chromatography for the enrichment of low
abundance protein
Affinity chromatography has recently been established as
a method for enriching proteins of low abundance. The tech-
nique is based on the specific interactions between immobi-
lized ligands and target proteins. These applications involve
the removal of a specific protein or a class of protein that
might interfere with 2-DE resolution enabling the visualiza-
tion of low abundance proteins in gel. The process is termed
‘immunoaffinity chromatography’ when an immune protein
is used as the ligand to target a specific protein. The technique
has successfully been used to visualize erythropoietin recep-
tor that is moderately expressed in cultured cell lines [28].
Under normal circumstances this protein cannot be visualized
in whole-cell extracts but can be enriched by antibody-based
affinity purification to yield a silver-stained band [28]. One of
the commonly used methods for phosphorylated proteins or
phosphopeptides is immobilized metal affinity chromatogra-
phy (IMAC) using gallium or iron-chelated affinity columns
that can selectively bind negatively charged phosphate groups
[29,30]. Affinity chromatography can only enrich a single low
abundance protein or a group of proteins at a time. There-
fore, multiple affinity columns and purification steps are re-
quired to achieve the desired result. On the other hand, using
group specific ligands such as heparin [31], triazine dye [32]
and metal ions (Cu2+ , Ni2+ and Zn2+ ) [33] one can separate
proteins into different groups. For example, using immobi-
lized metal ion affinity columns before 2-DE analysis, the
resolution of mouse liver proteins classified as metal bind-
ing and non-binding proteins was increased [33]. Similarly,
separated profiles of mitochondria calcium binding proteins,
glycoproteins and hydrophobic membrane proteins were ob-
tained by enriching these proteins using calcium chelated,
Con A and phenyl immobilized mini-spin affinity columns
fraction before 2-DE and MS analysis [34]. These studies
indicate that sub-cellular fractionation followed by affinity
chromatography based on non-specific interactions such as
ionic or hydrophobic interaction can be successfully used to
enrich a proteome for low abundance proteins of diversified
functional role.
3.6. Tracking low abundance cell signaling proteins
using proteomics approaches
Ten to fifteen percentage of the proteins encoded by the
human genome constitutes proteins involved in intracellu-
lar signaling. These proteins are expressed at a level that is
several hundred to thousand-fold lower than structural pro-
teins and proteins involved in metabolic pathways. This has
created an immense challenge for cell biologists to study
specific cell signaling proteins, either individually or in a
group. The difficulty in the analysis of signaling proteins can
be exemplified by the study of protein kinases involved in
the reversible phosphorylation of proteins, a major form of
post-translational events in cells. Nearly one third of pro-
teins inside the cells are subjected to such phosphorylation,
at multiple sites, approximately an average of 10 sites per
protein [35]. This permits a massive and complex cascade
of signaling networks inside the cell. The study of protein
phosphorylation by 2-DE methods has relied heavily on 32 P-
phosphate labeling of cells before resolution of proteins by
2-DE map. However, incubation of cells in culture in the pres-
 N. Ahmed, G.E. Rice / J. Chromatogr. B 815 (2005) 39–50
ence of millicurie amounts of 32 P-phosphate in low phosphate
containing medium to ensure sufficient labeling of intracel-
lular pool of ATP has been an issue. Both the high amount
of radioactive 32 P required for labeling and low phosphate
concentrations in the medium can induce stress response in
some cells, which can down regulate certain signaling path-
way enzymes making it difficult to evaluate the actions of
external stimuli. To study low abundant signaling events in
cells, some researchers have resorted to Western blot analy-
sis followed by 2-D PAGE. But this technique is completely
reliant on the availability of highly specific and potent an-
tibody that in many cases is not feasible. Some researchers
have resorted to the use of phosphosite specific antibodies
immobilized to metal affinity columns prior to 2-DE anal-
ysis [36,37]. However, to track signaling proteins for their
expression and activation states in high throughput fashion
it is necessary to develop affinity probes as specific as po-
tent antibodies that can be linked to solid support such as
beads, membranes of other matrices [35]. Many companies
are exploring the usage of such binding peptides to track sig-
naling proteins (e.g., Affibody, www.affibody.se). Recently,
the use of peptide antibody mimetics (PAM) has been intro-
duced by Kinexus Bioinformatics Corporation, in collabora-
tion with Biomime Corporation [35]. PAMs are short peptides
15–20 amino acids in length against a target protein directly
synthesized on a membrane. Usage of PAMs has the potential
of detecting low abundance signaling protein after 2-D PAGE.
Recent introduction of rapid affinity-capture of signal
transduction proteins (GRASP) has introduced a new phase in
molecular medicine by which the activity of signaling path-
ways from patient-derived carcinomas can be compared to
benign epithelial surfaces [38]. In this method, epithelium
from a carcinoma (benign or malignant) is scrapped without
the lysis of the basement membrane. Epithelial lysates are
prepared and the target protein is captured by immunopre-
cipitation and then resolved by 2-DE and analysed by West-
ern blotting [38]. This technology can be used to determine
the activation status of a signal transduction pathway by cap-
turing an appropriate target protein, either in an active or an
inactive state. GRASP represents an advance in proteomic
approaches as it detects protein–protein interaction present
in cells as they exist in vivo in their native tissue microenvi-
ronment.
3.7. Enrichment of low abundance hydrophobic and
basic proteins
Hydrophobic proteins are usually lost during sample
preparation or during entry of proteins into the IPG gels
or during focusing when the proteins reach their isoelectric
points. Recently, improvements in the solubilization of pro-
teins for the first dimension have been achieved by using more
effective reducing agents [39] and powerful chaotropes and
surfactants [39,40] than were previously used. Moreover, se-
quential extraction procedures that incorporate these solubi-
lizing agents and hydrophobic chromatographic techniques
43
have allowed the separation and detection of hydrophobic
proteins in cell lysates [41].
Sample pretreatment for the enrichment of basic protein
followed by focusing on narrow pI range strips beyond the
usual pH 10 endpoint of commercial IPG strips have allowed
the visualization and detection of basic proteins in cell lysates
[42]. The focusing of histones in a pH 10–12 IPG range has
successfully been shown [42]. Since then, the use of broad
range pH 3–12 and pH 4–12 IPGs have been shown to suc-
cessfully focus alkaline protein [43].
4. Strategies to enrich low abundance protein in
biological fluids
The introduction of new technologies for the detection of
disease specific biomarkers in the biological fluids of patients
will have an important impact on the health sector. This need
is particularly urgent in cancer and other diseases where early
diagnosis dramatically improves patient outcome [44].
Blood transports essential nutrients to the cells and carries
away metabolic waste products and other substances. Hence,
blood proteins are useful diagnostic tools and alteration of
the expression of some blood proteins is an early sign of an
altered physiology and may be indicative of disease [45]. Dur-
ing a heart attack or in other pathological conditions where
muscle degeneration is involved, damaged or dying cells se-
crete their contents into the bloodstream. The resulting con-
centration of one such serum protein, creatine-kinase B is a
measure of the amount of damaged muscle and is indicative
of a diseased state. In routine diagnostic laboratories identi-
fication of specific low abundant disease-associated proteins
in serum relies heavily on time consuming and expensive ra-
diolabeled or enzyme-linked immunoassay methods (RIA or
ELISA) that only have the ability to evaluate a single protein
component at a time. Due to the heterogenous nature of most
physiological disorders there is a common belief that no sin-
gle marker is likely to be sufficiently predictive [46] and a few
studies have emphasized the need for more than one candidate
biomarker to enhance already available diagnostic/prognostic
tests [47]. The necessity to develop a panel of multiple diag-
nostic/prognostic markers can be met by utilizing proteomic
approaches to plasma/serum and urine specimens that have
the capacity to profile multiple biomarkers [48,49]. How-
ever, the application has been limited by the presence of high
abundance ‘common housekeeping’ proteins like albumin,
immunoglobulins, transferrin, haptoglobin, etc. These pro-
teins constitute approximately 60–97% of the total serum
protein. If proteomics technologies are to be used routinely
for diagnostic purposes, a rapid, inexpensive and a simple
method is required to remove these high abundance proteins.
4.1. Application of affinity chromatograph in biological
fluids
For the affinity depletion of high abundance protein in bi-
ological fluids, the technique of ‘affinity extraction’ can be
44 N. Ahmed, G.E. Rice / J. Chromatogr. B 815 (2005) 39–50
applied to deplete a biological fluid sample of a specific pro-
tein or a group of proteins from a sample before analysis with
2-DE method. Examples include the use of immobilized pro-
tein A and anti-immunoglobulin support to adsorb selectively
immune complexes from patient samples [50]. This method
can be used in conjunction with the depletion of albumin
from biological fluid samples such as serum, peritoneal fluid
or cerebospinal fluid samples by adsorption on affinity resins
[51]. Recently, ProtoClear, a patented affinity matrix, based
on the small molecule Cibacron Blue has been used to remove
albumin and IgG from human serum samples. This led to im-
proved resolution of protein spots in 2-DE gel maps [52].
Recently, we have shown the use of Affi-Gel Blue (another
cibacron blue based matrix, Bio-Rad laboratories) and pro-
tein A in the depletion of albumin from human serum [51].
This method resulted in the removal of proteins co-migrating
at 68 kDa with concomitant visualization of 28 unique spots
and enhanced visualization of several common protein spots
after Affi-Gel Blue treatment (Fig. 2). A time course stud-
ies on human serum after Affi-Gel Blue treatment showed
concomitant removal of proteins other than albumin (Fig. 2).
Within 1 h of Affi-Gel Blue treatment, significant loss of al-
bumin was achieved with no significant loss of protein profile
or number of protein spots [51]. However, 16 h treatment of
Affi-Gel Blue resulted in the loss of albumin with approxi-
mately 35% loss of total number of protein spots compared
to 1 h treatment [51] (Fig. 1). These observations suggest
that even though 16 h exposure of human serum to Affi-Gel
Blue can result in significant loss of albumin and consequent
enhancement of several low abundance proteins, it is also
associated with nonspecific removal of serum protein other
than albumin. We have recently shown enhanced visualiza-
tion of potential biomarkers in the serum of ovarian cancer
patients using 1 h Affi-Gel Blue treatment method [53]. Other
Fig. 1. Time dependent removal of albumin after treatment of human serum with Affi-Gel Blue. Serum sample was treated with Affi-Gel Blue for 0 min, 1 and
16 h before being resolved by 2-D PAGE.
dye-ligands that could alternatively be employed to remove
abundant blood proteins include Procion Red He3B, Reactive
Blue MRB, Reactive Green H4G, Reactive Green HE4BD,
Reactive Yellow M8G and Reactive Brown M4R all of which
can be coupled to supports such as Sepharose 4B or 6B [54].
However, the draw back in the usage of these reactive dyes
is that the planar ring structure of these dyes through a com-
plex combination of electrostatic, hydrophobic and hydrogen
bonding can bind other proteins besides albumin resulting in
non-specific removal of proteins [51,52]. Hence, the use of
reactive dyes in affinity based depletion methods for the re-
moval of high abundance proteins should be used with cau-
tion. However, under optimized conditions the use of reactive
affinity dyes in conjunction with a supporting matrix can be
a useful tool for the depletion of high abundant proteins from
biological fluids [51,52].
In an effort to remove high abundance protein from cere-
brospinal fluid to identify markers for Alzheimer’s disease,
Patterson [55] in combination with Oxford GlycoSciences
and Pfizer have developed a method for selective removal of
high abundant albumin, IgG, transferrin and haptoglobin by
affinity depletion before 2-DE analysis. The study involved
512 samples and resulted in the visualization of 1131 protein
spots in 2-DE map and the identification of potential markers
of Alzheimer’s disease [55]. Recently, Agilent Technologies
have introduced “Agilent multiple affinity removal system”
that binds and retain six highly abundant protein-albumin,
IgG, IgA, transferrin, antitrypsin and haptoglobin in one step
(Agilent Technologies Inc., USA). This multiple affinity re-
moval system uses 1100 series HPLC to ensure consistency of
results within the samples and the ability to automate sample
processing. Hence, the technology combines the specificity of
antibody–antigen recognition with the efficiency of standard
liquid chromatography instrumentation resulting in substan-
 N. Ahmed, G.E. Rice / J. Chromatogr. B 815 (2005) 39–50
tial removal of high abundance protein with subsequent in-
creased loading of low-abundance proteins onto gels and en-
hanced detection of low-abundance serum proteins in a high
throughput fashion. In a similar fashion, the minor proteins
in human milk were analyzed by using immuno-adsorbents
to remove three major proteins (␣-lactalbumin, lactoferrin
and secretory immunoglobulin A) [56]. In another study, an
immunoaffinity column that contained ␤-casein and bovine
IgG-specific immobilized sepharose was used to remove
major proteins and low abundance bovine milk proteins were
identified [57]. In addition to these affinity methods, an affin-
ity spin tube filter method that relies on protein G-antibody
binding has been used either to remove high abundance
protein or to enrich the proteome of specific low abundance
proteins [58]. These studies indicate that immunoaffinity
based methods are effective in selective removal of high
abundance proteins and are gaining considerable importance
for the identification of low abundance proteins in biological
fluids.
4.2. High abundant protein depletion strategies using
multiple chicken IgY antibodies
Recently, the use of chicken IgY antibodies raised against
high abundant proteins such as albumin, transferrin, fibrino-
gen, IgA, IgM, IgG covalently linked to a solid support
has gained prominence in the depletion of human plasma
of high abundance protein before 2-DE PAGE (Charles
River Proteomic Services, Worcester, MA). This method al-
lows high throughput depletion of abundant proteins from
serum/plasma of multiple mammalian species. Due to greater
evolutionary differences between birds and mammals, there is
a greater probability of producing specific and potent antibod-
ies against evolutionary conserved high abundant mammalian
antigens by immunizing chickens. The yolk of eggs laid by
an immunizied chicken is an abundant source of polyclonal
antibodies. The avian equivalent of IgG, usually referred as
IgY, is significantly different from IgG. In addition to having
different sequence, IgY has higher molecular weight, higher
electrophoretic mobility and lower isoelectric pH. Moreover,
ionic detergent has no effect on the affinity of IgY for dif-
ferent antigens but can inhibit the interaction of IgG with
some antigens. The hinge region present between the Fab
pieces of the immunoglobulin is absent in IgY. This hinge re-
gion renders IgG less stable than IgY, making IgY more suit-
able to be covalently linked to solid-phase support. Moreover,
chicken IgY does not cross react with mammalian IgG nor
does it bind bacterial or mammalian receptors, reducing non-
specific binding and eliminating the need for cross-species
immunoadsorptions.
Recently, we have been successful in developing ‘first
and second generation chicken antibodies’ capable of de-
pleting human plasma of high abundance protein (Fig. 2,
Rice et al., unpublished data). The ‘first generation anti-
body’ was raised by immunizing chicken with multiple doses
of whole plasma. The ‘second-generation antibody’ on the
45
other hand, was raised against plasma that had already been
subjected to affinity depletion by Affi-Gel Blue and immun-
odepletion using chicken IgY obtained from the first round
(Fig. 3). Such method of immunodepletion before 2-D PAGE
drastically reduced the visualization of high abundance pro-
tein making it possible to detect protein of low abundance
(Fig. 3).
4.3. Low abundance protein enrichment in biological
fluids by sample pre-fractionation
Pre-fractionation of a protein pool prior to 2-DE can create
discrete sample pools allowing better separation and identifi-
cation of low abundance proteins. Current pre-fractionation
methods include sequential extraction with increasingly
stronger solubilization solution [59,60], sub-cellular frac-
tionation [61,62], and selective removal of high abundance
protein by using affinity chromatography methods. Recently,
pre-fraction methods based on isoelectric trapping methods
have been developed. The method relies on the use of multi-
compartment electrolyzer with an isoelectric membrane that
separates plasma proteins into three main pI fractions, an
acidic fraction, a basic fraction and an albumin fraction with a
pI between 5.6 and 6.1 [27]. This method of fractionation pro-
cedure enables increased protein loads and eliminates protein
precipitation during IEF and increases the number of more
acidic proteins visualized on narrow range 2-DE gels com-
pared to un-fractionated plasma [27]. However, major draw
back in this method is the loss of protein close to the pI 5.6 due
to protein precipitation on the membrane. Others have used
a miniaturized form of an IEF device or a flat bed granulated
Sephadex gel [63] or SDS–PAGE-based size fractionation
techniques [64] to enrich biological fluids of low abundance
proteins.
Gradiflow method that relies on protein separation based
on molecular size by selection of specific separation mem-
brane cut off size and charge on the protein by adjusting the
pH of the system has also been used [65]. Horvath et al. [66]
reported the use of cibacron blue affinity matrices in the Grad-
iflow system to enrich the majority of serum proteins through
the depletion of albumin. While others have reported the com-
bination of only charge and size separation capacity of Grad-
iflow to deplete plasma of albumin [66]. Using the Gradiflow
method, one can separate proteins in a biological fluid at high
ionic strength to help maintain the solubility of proteins near
their isoelectric point without the use of denaturants [65]. As
the method does not use IEF membranes for the charge sepa-
ration of proteins, protein aggregation and precipitation com-
mon in conventional IEF-based membrane separation system
can be minimized. On the other hand, Gradiflow’s ability to
separate proteins by molecular weight enables the separation
of serum proteins away from albumin.
The abundant profile of human urine changes as a re-
sult of disease state or drug toxicity. To succeed in iden-
tifying the differential protein in human urinary proteome,
protein fractionation strategies enriching proteins of molec-
46 N. Ahmed, G.E. Rice / J. Chromatogr. B 815 (2005) 39–50

Fig. 2. Schematic representation of a device for the processing of multiple chicken IgY antibodies.

ular masses lower than 30 kDa is required as the kidney re-
stricts passage of plasma proteins from the glomerular fil-
trate in the Mr range above 40 kDa. Pieper et al. [67] has
recently described a protein fractionation strategy separat-
ing proteins of molecular mass lower than 30 kDa from
larger proteins. The fraction containing proteins with Mr
higher than 30 kDa was depleted of high abundant albumin
and immunoglobulin G by immunoaffinity subtraction chro-
matography [67]. Following 2-DE display, superior protein
resolution was obtained and the application of mass spec-
trophotometry to protein spots led to a successful identifica-
tion of 30% of urinary proteins with Mr values higher than
30 kDa. The fractionation approach used by Piper et al. [67]
holds promise for the biomarker discovery of proteins from
the urine of patients suffering from different pathological
conditions.
4.4. Use of overlapping narrower pI range strips for
enrichment of low abundance protein in biological fluids
The use of wide range pH gradient such as 4–7 or 3–11
gives an overview profile of a proteome displaying hundreds
of overlapping proteins that are impossible to detect and iden-
tify. In order to detect proteins of low abundance it is crucial
to simplify such profiling of protein so that individual spot
can be visualized. The advantage of using overlapping nar-
row immobilized pH gradients is to gain higher resolution of
protein by stretching the protein pattern in the first dimen-
sion. This simplifies the computer aided image analysis and
allows the visualization and identification of more proteins.
This method has been used for the resolution of a yeast pro-
teome where a total of 2286 yeast proteins were spotted using
pH gradients of 4–5, 4.5–5.5, 5–6, 5.5–6.7 and 6.9 compared
 N. Ahmed, G.E. Rice / J. Chromatogr. B 815 (2005) 39–50 47

Fig. 3. 2-DE profile of human serum before and after treatment with chicken IgY antibodies: (a) profile of untreated human serum; (b) display of proteins
present after Affi-Gel Blue treatment and (c) profile of proteins present after treatment with Affi-Gel Blue and then anti-human serum Sepharose 4B. Differences
in the protein profile are indicated by blue, red and yellow colour. In neat untreated serum (a) protein smear outlined in red is albumin.

to 755 protein spots in the pH 3–10 gradient [68]. Hoving
et al. [69] have also used this method to visualize a larger
proportion of the B-lymphoma proteome to detect proteins
present at low copy numbers. Using such over lapping narrow
pI strips recently we have been successful in identifying low
abundant biomarkers in serum from ovarian cancer patients
(Ahmed et al., unpublished data). However, overlapping nar-
rower pI range strips can be used successfully for acidic and
neutral pI proteins. Protein separation in the alkaline pH range
has recently been optimized [69].
4.5. Enrichment of low abundance protein by localized
venous drainage
One of the methods that has evoked recent interest for pro-
tein expression profiling for markers of cancer involves col-
lection of venous blood close to the site of tumor growth. Such
localized drainage of venous blood allows analysis of concen-
trated protein mixture without being diluted by the systemic
circulation. The underlying rationale behind this approach is
that the cells in tumor tissues are likely to over express pro-
teins that may be released directly or shed by ‘ectodomain
shedding’ a process, in which the extracellular domain of
transmembrane proteins are proteolytically cleaved from the
cell surface and shed into the neighboring blood vessels be-
fore being diluted by the systemic circulation. For example,
the cancer specific antigen 125 (CA-125) is a serum marker
for the diagnosis of ovarian cancer. CA-125 shows wide vari-
ation in sensitivity and specificity in the serum of ovarian
cancer patient [70], yet high levels of this marker may be
detected in ovarian cancer patient’s ovarian venous drainage.
Due to the multi-factorial genesis of cancer, it is likely that a
combination of several markers will be required to effectively
predict the biological behavior of a tumor. For example, neo-
plasms are classified into several subtypes and are divided
into primary and metastases tumors originating from the pri-
mary site. Hence, in cancer research there is a need to develop
markers that will aid in the classification of these different tu-
mors. This is not only required for academic interest but is
also crucial for optimal treatment choices. In this context, use
of venous blood close to the tumor site may serve as a useful
source for the development of markers that may aid in the
classification of different tumors.
Apart from the low abundant protein enriching fractiona-
tion methods described above, protein–protein interaction in
immunoglobulin G-associated minor proteins in biological
fluids can be visualized and identified by techniques that em-
ploy combination of denaturing and nondenaturing 2-DE gels
(1% agarose) [71]. These methods allow the visualization and
identification of native protein non-covalently associated in
48 N. Ahmed, G.E. Rice / J. Chromatogr. B 815 (2005) 39–50
complexes. Structural analyses of these proteins may provide
new aspect on the functions of proteins present in biological
fluids.
5. Advances and limitations of 2-DE based
technology
A number of methodological improvements have been
made to 2-DE technology since its introduction by O’Farrell
and Klose in 1975. The development of immobilized pH
gradients of different ranges has made the technique more
reproducible and allowed comparison of results between
inter-laboratory groups. This improvement has also resulted
in increasing the loading capacity of a sample making the
technique more sensitive for the detection and identifi-
cation of low abundance protein. The resolution of basic
and hydrophobic proteins has also been increased by the
introduction of narrower basic pH strips [69] and by the intro-
duction of new reagents such as more efficient detergents and
chaotropes. The use of novel extraction procedures to deplete
a complex proteome of high abundance proteins present in
several orders of magnitude have made possible the visualiza-
tion and detection of low abundance biologically important
proteins. However, the extraction procedures currently in use
result in the several-fold dilution of protein samples leaving
one with inadequate concentration of protein to load on the
gel. Hence, to achieve increased protein loading capacity of
a proteome subjected to high abundance protein depletion
procedures, optimal method of concentrating dilute protein
samples is required. The use of precipitation techniques,
high molecular weight cutoff centrifugal filters and freeze
drying devices has greatly facilitated in concentrating dilute
protein mixtures over hundred fold and have thus, aided in
overcoming the protein loading problem.
The introduction of some of the more sensitive staining
dyes has increased the visualization and made identification
of protein spots present in minute amount feasible. Previ-
ously, visualization of proteins separated by 2-D PAGE was
dependent on Coomassie blue or silver staining. However,
the recent introduction of fluorescent dyes offers substantial
advantage over these colorimetric detection methods and has
increased the sensitivity and quantitative reproducibility of a
proteomic profile over a broad dynamic range. Fluorescence
two-dimensional differential gel electrophoresis (2-D DIGE)
is a new advancement in proteomics technology. In this tech-
nique, proteins can be labeled co-valently by two fluorescent
dyes prior to isoelectric focusing. The mixture of protein is
then run on a single 2-DE gel. This method allows differential
expression analysis of protein and significantly reduces the
artifact problem introduced in gel matching using standard
fluorescent dyes such as Sypro Ruby.
One area within the field of proteomics that has ex-
panded most is the automation of protein characteriza-
tion in mass spectrometry (MS). With full automation,
the matrix assisted laser desorption and ionization time
of flight mass spectrometry (MALDI-TOF-MS) and nano-
electrospray quadrupole-quadrupole time of flight mass spec-
trometry (n-ESIQ(q)TOF-MS) is capable of analyzing sev-
eral hundreds of proteins separated on a gel within a reason-
able time period. The introduction of robotic spot cutters and
robot guided image analysis software results in high through-
put identification of proteins separated on a gel. The improved
sensitivity of the MS instrument at present can identify pro-
teins present in picomole quantity [72].
The development in computer-based image analysis sys-
tems (e.g., PDQUEST and MELANIE) has allowed the ef-
ficient evaluation of hundreds of proteins separated on 2-
DE gels. However, the accuracy of gel matching can decline
with the heterogenity between samples being compared and
if the experiments are run under different conditions. Even
though, most of the available 2-DE software has the au-
tomated processes of spot detection, quantification, match-
ing and statistical analysis, a limited level of manual editing
and re-evaluation is still required. Hence, caution should be
taken during manual editing and re-evaluation to eliminate
the chances of introducing false positives.
The availability of public databases has made the iden-
tification of sequenced proteins easy. These databases are
regularly updated and offer access to information on differ-
ent proteins. These databases can be reached through Web
sites like ExPASy server in Geneva or NCBI port in USA.
The availability of such databases makes inter-laboratory gel
comparisons possible and enhances protein spot identifica-
tion by gel matching. However, exchange of 2-DE gel pat-
terns between inter-laboratory experiments should be treated
with caution as minor variations in sample handling before 2-
DE may give rise to protein patterns with specific loss or gain
of protein spots. Interesting developments in proteome data
acquisition that incorporate the development of laboratory
information processing systems (LIPS) [73] such as sam-
ple information, clinical diagnosis, analyzed images, protein
identification and quantification, etc will increase the possi-
bilities of multivariate analysis of data and will prevent the
loss of vital biological information. Such an integrated net-
work of information will result in clinicopathological diag-
nosis of disease resulting in accurate prognosis and treatment
prediction.
6. Future perspectives and conclusions
The focus in biomedical research is to identify early stage
markers for the diagnosis and better therapeutic treatment
of diseases or to determine the molecular defect that under-
lines a specific disease or to identify potential targets for drug
treatment. Proteomics is used as a technology to solve the
underlying basis of all these problems. The basis of the tech-
nology is not only to list differentially expressed proteins in
a proteome, but the technology acts as a circuit where the in-
tercellular information generated in a proteome is portrayed
in a protein profile. The differences in the protein profile
 N. Ahmed, G.E. Rice / J. Chromatogr. B 815 (2005) 39–50
of a diseased proteome compared to that of a normal one
could be potentially used as a diagnostic marker. However,
the search for a single marker for disease identification or a
single target for drug treatment is not likely to be success-
ful for complex diseases such as cancer. Proteomics is the
only technology that offers sufficient resolutions to under-
stand the subtle phenotype changes in a diseased proteome
with a view to develop better markers and drug targets. How-
ever, there is a great need for the development of methods
that would simplify complex protein mixtures of a human
proteome. The technology is challenged by the presence of
as many as 100,000 proteins with a dynamic range estimated
between 5 and 12 orders of magnitude (7). Scientists working
with human cells have long recognized the need to develop
pre-fractionation procedures either at the sub-cellular level
or in combination with isoelectric fractionation to deplete the
human proteome of abundant proteins so that successful min-
ing for low abundant proteins is achievable. The intense on-
going effort addresses the development of several advanced
physicochemical methods for protein fractionation. The use
of sub-cellular fractionation protocols, affinity methods, laser
micro-dissection technology, fractionation by narrower pI
range strips and antibody adsorption methods have gained
recent interest. In most cases, the use of a single fractiona-
tion method is not sensitive to detect the rare low abundant
proteins and a combination of sub-cellular and iso-electric
fractionation protocols are required to allow quantitative as-
sessment of the enriched minor components. By allowing
one to adhere to such multiple fractionation methods there is
a possibility that the proteins of interest may show up in one
or more fractions. With the development of each fractionation
protocol for tracking low abundant proteins new information
is being obtained. Procurement of such information is es-
sential for biological/clinical researchers to understand the
complex biological processes governing a diseased state and
to make progress in the diagnostic and prognostic fronts. But
before that can be achieved, a period of exploration and opti-
mization is still required to simplify the human proteome by
high abundant protein depletion.
Acknowledgements
The authors wish to acknowledge the technical help of Ms.
Karen Oliva and Ms. Gillian Barker. The proteomics study
is supported by the Jack Brockhoff Foundation, the Cancer
Council of Victoria, the Rotary Club of Williamstown, Ov-
care, Jigsaw Women’s Fashion Company, B.O.O.T.S. ALL
Inc. (Breasts Ovaries and other things sacred) and BHP Bil-
liton Community Trust Australia.
References
[1] J.C. Venter, M.D. Adams, E.W. Myers, P.W. Li, R.J. Mural, G.G.
Sutton, H.O. Smith, M. Yandell, C.A. Evans, R.A. Holt, J.D. Go-
cayne, P. Amanatides, R.M. Ballew, D.H. Huson, J.R. Wortman, Q.
49
Zhang, C.D. Kodira, X.H. Zheng, L. Chen, M. Skupski, G. Sub-
ramanian, P.D. Thomas, J. Zhang, G.L. Gabor Miklos, C. Nelson,
S. Broder, A.G. Clark, J. Nadeau, V.A. McKusick, N. Zinder, A.J.
Levine, R.J. Roberts, M. Simon, C. Slayman, M. Hunkapiller, R.
Bolanos, A. Delcher, I. Dew, D. Fasulo, M. Flanigan, L. Florea,
A. Halpern, S. Hannenhalli, S. Kravitz, S. Levy, C. Mobarry, K.
Reinert, K. Remington, J. Abu-Threideh, E. Beasley, K. Biddick, V.
Bonazzi, R. Brandon, M. Cargill, I. Chandramouliswaran, R. Char-
lab, K. Chaturvedi, Z. Deng, V. Di Francesco, P. Dunn, K. Eilbeck,
C. Evangelista, A.E. Gabrielian, W. Gan, W. Ge, F. Gong, Z. Gu,
P. Guan, T.J. Heiman, M.E. Higgins, R.R. Ji, Z. Ke, K.A. Ketchum,
Z. Lai, Y. Lei, Z. Li, J. Li, Y. Liang, X. Lin, F. Lu, G.V. Merkulov,
N. Milshina, H.M. Moore, A.K. Naik, V.A. Narayan, B. Neelam, D.
Nusskern, D.B. Rusch, S. Salzberg, W. Shao, B. Shue, J. Sun, Z.
Wang, A. Wang, X. Wang, J. Wang, M. Wei, R. Wides, C. Xiao, C.
Yan, et al., Science 291 (2001) 1304.
[2] M. Labrador, F. Mongelard, P. Plata-Rengifo, E.M. Baxter, V.G.
Corces, T.I. Gerasimova, Nature 409 (2001) 1000.
[3] M.R. Wilkins, J.C. Sanchez, A.A. Gooley, R.D. Appel, I. Humphery-
Smith, D.F. Hochstrasser, K.L. Williams, Biotechnol. Genet. Eng.
Rev. 13 (1996) 19.
[4] M.R. Wilkins, J.C. Sanchez, K.L. Williams, D.F. Hochstrasser, Elec-
trophoresis 17 (1996) 830.
[5] V.C. Wasinger, S.J. Cordwell, A. Cerpa-Poljak, J.X. Yan, A.A.
Gooley, M.R. Wilkins, M.W. Duncan, R. Harris, K.L. Williams, I.
Humphery-Smith, Electrophoresis 16 (1995) 1090.
[6] R.A. VanBogelen, K.Z. Abshire, B. Moldover, E.R. Olson, F.C. Nei-
dhardt, Electrophoresis 18 (1997) 1243.
[7] G.L. Corthals, V.C. Wasinger, D.F. Hochstrasser, J.C. Sanchez, Elec-
trophoresis 21 (2000) 1104.
[8] B. Futcher, G.I. Latter, P. Monardo, C.S. McLaughlin, J.I. Garrels,
Mol. Cell Biol. 19 (1999) 7357.
[9] B. Franzen, T. Hirano, K. Okuzawa, K. Uryu, A.A. Alaiya, S. Linder,
G. Auer, H. Kato, Electrophoresis 16 (1995) 1087.
[10] B. Franzen, S. Linder, K. Okuzawa, H. Kato, G. Auer, Electrophore-
sis 14 (1993) 1045.
[11] M.A. Reymond, J.C. Sanchez, G.J. Hughes, K. Gunther, J. Riese,
S. Tortola, M.A. Peinado, T. Kirchner, W. Hohenberger, D.F.
Hochstrasser, F. Kockerling, Electrophoresis 18 (1997) 2842.
[12] C. Sarto, A. Marocchi, J.C. Sanchez, D. Giannone, S. Frutiger, O.
Golaz, M.R. Wilkins, G. Doro, F. Cappellano, G. Hughes, D.F.
Hochstrasser, P. Mocarelli, Electrophoresis 18 (1997) 599.
[13] L. Bini, B. Magi, B. Marzocchi, F. Arcuri, S. Tripodi, M. Cintorino,
J.C. Sanchez, S. Frutiger, G. Hughes, V. Pallini, D.F. Hochstrasser,
P. Tosi, Electrophoresis 18 (1997) 2832.
[14] A. Okada, J.P. Bellocq, N. Rouyer, M.P. Chenard, M.C. Rio, P.
Chambon, P. Basset, Proc. Natl. Acad. Sci. U.S.A. 92 (1995)
2730.
[15] S.P. Gygi, R. Aebersold, Methods Mol. Biol. 112 (1999) 417.
[16] L. Liotta, E. Petricoin, N.L. Simone, C.P. Paweletz, L. Charboneau,
E.F. Petricoin III, L.A. Liotta, Nat. Rev. Genet. 1 (2000) 48.
[17] Y. Zheng, Y. Xu, B. Ye, J. Lei, M.H. Weinstein, M.P. O’Leary, J.P.
Richie, S.C. Mok, B.C. Liu, Cancer 98 (2003) 2576.
[18] R.E. Banks, M.J. Dunn, M.A. Forbes, A. Stanley, D. Pappin, T.
Naven, M. Gough, P. Harnden, P.J. Selby, Electrophoresis 20 (1999)
689.
[19] A.P. Fuller, D. Palmer-Toy, M.G. Erlander, D.C. Sgroi, J. Mammary
Gland Biol. Neoplasia 8 (2003) 335.
[20] J.D. Wulfkuhle, J.A. Aquino, V.S. Calvert, D.A. Fishman, G. Coukos,
L.A. Liotta, E.F. Petricoin III, M.B. Jones, H. Krutzsch, H. Shu, Y.
Zhao, E.C. Kohn, Proteomics 3 (2003) 2085.
[21] M.B. Jones, H. Krutzsch, H. Shu, Y. Zhao, L.A. Liotta, E.C. Kohn,
E.F. Petricoin III, Proteomics 2 (2002) 76.
[22] F. Zanconati, F. Martellani, D. Bonifacio, G. Falanga, L. Di Bonito,
Pathologica 95 (2003) 257.
[23] C. Uras, E. Altinkaya, H. Yardimci, S. Goksel, N. Yavuz, L. Kap-
tanoglu, T. Akcal, Surg. Oncol. 5 (1996) 259.
50 N. Ahmed, G.E. Rice / J. Chromatogr. B 815 (2005) 39–50
[24] S. Snape, R.J. Simpson, T.J. Peters, Cell Biochem. Funct. 8 (1990)
107.
[25] M. Fountoulakis, B. Takacs, Methods Enzymol. 358 (2002) 288.
[26] M. Fountoulakis, J.F. Juranville, Anal. Biochem. 313 (2003) 267.
[27] B. Herbert, P.G. Righetti, Electrophoresis 21 (2000) 3639.
[28] A. Pandey, M. Mann, Nature 405 (2000) 837.
[29] M.C. Posewitz, P. Tempst, Anal. Chem. 71 (1999) 2883.
[30] M. Mann, R.C. Hendrickson, A. Pandey, Annu. Rev. Biochem. 70
(2001) 437.
[31] P. Jungblut, J. Klose, Biochem. Genet. 24 (1986) 925.
[32] P. Jungblut, J. Klose, J. Chromatogr. 482 (1989) 125.
[33] P. Jungblut, H. Baumeister, J. Klose, Electrophoresis 14 (1993) 638.
[34] M.F. Lopez, B.S. Kristal, E. Chernokalskaya, A. Lazarev, A.I.
Shestopalov, A. Bogdanova, M. Robinson, Electrophoresis 21 (2000)
3427.
[35] S. Pelech, Curr. Pharm. Biotechnol. 5 (2004) 69.
[36] P.B. Maguire, K.J. Wynne, D.F. Harney, N.M. O’Donoghue, G.
Stephens, D.J. Fitzgerald, Proteomics 2 (2002) 642.
[37] M. Gronborg, T.Z. Kristiansen, A. Stensballe, J.S. Andersen, O.
Ohara, M. Mann, O.N. Jensen, A. Pandey, Mol. Cell Proteomics
1 (2002) 517.
[38] D.M. Berman, I.M. Shih, L.A. Burke, T.D. Veenstra, Y. Zhao, T.P.
Conrads, S.W. Kwon, V. Hoang, L.R. Yu, M. Zhou, R.J. Kurman,
E.F. Petricoin, L.A. Liotta, Proteomics 4 (2004) 812–818.
[39] B.R. Herbert, M.P. Molloy, A.A. Gooley, B.J. Walsh, W.G. Bryson,
K.L. Williams, Electrophoresis 19 (1998) 845.
[40] M. Fountoulakis, B. Takacs, Electrophoresis 22 (2001) 1593.
[41] M. Fountoulakis, M.F. Takacs, B. Takacs, J. Chromatogr. A 833
(1999) 157.
[42] P.G. Righetti, A. Bossi, A. Gorg, C. Obermaier, G. Boguth, J.
Biochem. Biophys. Methods 31 (1996) 81.
[43] A. Gorg, C. Obermaier, G. Boguth, W. Weiss, Electrophoresis 20
(1999) 712.
[44] E.F. Petricoin, A.M. Ardekani, B.A. Hitt, P.J. Levine, V.A. Fusaro,
S.M. Steinberg, G.B. Mills, C. Simone, D.A. Fishman, E.C. Kohn,
L.A. Liotta, Lancet 359 (2002) 572.
[45] T.C. Poon, P.J. Johnson, Clin. Chim. Acta 313 (2001) 231.
[46] R.C. Bast Jr., N. Urban, V. Shridhar, D. Smith, Z. Zhang, S. Skates,
K. Lu, J. Liu, D. Fishman, G. Mills, Cancer Treat. Res. 107 (2002)
61.
[47] B.L. Adam, A. Vlahou, O.J. Semmes, G.L. Wright Jr., Proteomics
1 (2001) 1264.
[48] M.B. Daly, R.F. Ozols, Cancer Cell 1 (2002) 111.
[49] R. Pieper, C.L. Gatlin, A.M. McGrath, A.J. Makusky, M. Mondal,
M. Seonarain, E. Field, C.R. Schatz, M.A. Estock, N. Ahmed, N.G.
Anderson, S. Steiner, Proteomics 4 (2004) 1159.
[50] D.S. Hage, Clin. Chem. 45 (1999) 593.
[51] N. Ahmed, G. Barker, K. Oliva, D. Garfin, K. Talmadge, H. Geor-
giou, M. Quinn, G. Rice, Proteomics 3 (2003) 1980.
[52] B.A. Lollo, S. Harvey, J. Liao, A.C. Stevens, R. Wagenknecht, R.
Sayen, J. Whaley, F.G. Sajjadi, Electrophoresis 20 (1999) 854.
[53] N. Ahmed, G. Barker, K.T. Oliva, P. Hoffmann, C. Riley, S. Reeve,
A.I. Smith, B.E. Kemp, M.A. Quinn, G.E. Rice, Br. J. Cancer 91
(2004) 129.
[54] A.F. McGettrick, D.M. Worrall, Methods Mol. Biol. 244 (2004)
151.
[55] S.D. Patterson, Physiol. Genomics 2 (2000) 59.
[56] K. Murakami, M. Lagarde, Y. Yuki, Electrophoresis 19 (1998) 2521.
[57] M. Yamada, K. Murakami, J.C. Wallingford, Y. Yuki, Electrophoresis
23 (2002) 1153.
[58] Y.Y. Wang, P. Cheng, D.W. Chan, Proteomics 3 (2003) 243.
[59] M.P. Molloy, B.R. Herbert, K.L. Williams, A.A. Gooley, Elec-
trophoresis 20 (1999) 701.
[60] M.P. Molloy, B.R. Herbert, M.B. Slade, T. Rabilloud, A.S. Nouwens,
K.L. Williams, A.A. Gooley, Eur. J. Biochem. 267 (2000) 2871.
[61] C. Pasquali, I. Fialka, L.A. Huber, F. Lottspeich, H. Ahorn, J. Chro-
matogr. B Biomed. Sci. Appl. 722 (1999) 89.
[62] I. Fialka, C. Pasquali, F. Lottspeich, H. Ahorn, L.A. Huber, Elec-
trophoresis 18 (1997) 2582.
[63] A. Gorg, G. Boguth, A. Kopf, G. Reil, H. Parlar, W. Weiss, Pro-
teomics 2 (2002) 1652.
[64] S.T. Kim, H.S. Kim, H.J. Kim, S.G. Kim, S.Y. Kang, D.B. Lim,
K.Y. Kang, Mol. Cells 16 (2003) 316.
[65] D.L. Rothemund, V.L. Locke, A. Liew, T.M. Thomas, V. Wasinger,
D.B. Rylatt, Proteomics 3 (2003) 279.
[66] S.Z. Hovarth, A.A. Gooley, C.W. Wrigley, J. Margolis, K.L.
Williams, Electrophoresis 17 (1996) 224.
[67] R. Pieper, C.L. Gatlin, A.M. McGrath, A.J. Makusky, M. Mondal,
M. Seonarain, E. Field, C.R. Schatz, M.A. Estock, N. Ahmed, N.G.
Anderson, S. Steiner, Proteomics 4 (2004) 1159.
[68] R. Wildgruber, A. Harder, C. Obermaier, G. Boguth, W. Weiss, S.J.
Fey, P.M. Larsen, A. Gorg, Electrophoresis 21 (2000) 2610.
[69] S. Hoving, H. Voshol, J. van Oostrum, Electrophoresis 21 (2000)
2617.
[70] B. Bastani, N. Chu, Am. J. Nephrol. 15 (1995) 468.
[71] T. Manabe, N. Yamaguchi, J. Mukai, O. Hamada, O. Tani, Pro-
teomics 3 (2003) 832.
[72] A. Shevchenko, M. Wilm, O. Vorm, O.N. Jensen, A.V. Podtelejnikov,
G. Neubauer, P. Mortensen, M. Mann, Biochem. Soc. Trans. 24
(1996) 893.
[73] I. Ali, Y. Chan, R. Kuick, D. Teichroew, S.M. Hanash, Electrophore-
sis 12 (1991) 747.